EphA2/Ephrin-A1 Signaling Complexes Restrict Corneal Epithelial Cell Migration

Abstract

Purpose.: Eph/ephrin signaling proteins are present in the corneal epithelium, where their function remains unknown. The authors examined the role of the EphA2 receptor and ephrin-A1 ligand in human corneal epithelial cell migration.

Methods.: Immunohistochemical analysis of EphA2 and ephrin-A1 in healthy and diabetic corneas was performed in concert with linear scratch wound healing studies in primary and telomerase-immortalized human corneal epithelial cells. Corneal epithelial cells were exposed to a soluble ephrin-A1-Fc peptide mimetic that targets EphA2 to trigger receptor phosphorylation and subsequent downregulation. Genetic modulation of EphA2 and ephrin-A1 levels was combined with manipulation of Erk1/2 or Akt signaling during wound healing.

Results.: EphA2 was immunolocalized to human corneal epithelial cells in vivo and in vitro. Ephrin-A1 ligand targeting of EphA2 restricted the ability of corneal epithelial cells to seal linear scratch wounds in a manner that was associated with a transient reduction in Erk1/2 and Akt activation state. Ephrin-A1-Fc treatment delayed wound healing independently of Mek-Erk1/2 signaling but was no longer capable of restricting migration after pharmacologic blockade of the PI3K-Akt pathway. Interestingly, ephrin-A1 immunoreactivity was increased in the corneal epithelia of diabetic individuals, mice maintained on a high-fat diet, or cultured corneal epithelial cells exposed to high glucose, which exhibit impaired Akt signaling and slower wound healing responses.

Conclusions.: EphA2 attenuates corneal epithelial cell migration when stimulated by ephrin-A1 ligand in a manner that involves the suppression of Akt. Elevated levels of ephrin-A1 may contribute to diabetic keratopathies by persistently engaging EphA2 and prohibiting Akt-dependent corneal epithelial repair processes.

The cornea is lined by a stratified, nonkeratinizing epithelium that, in conjunction with the tear film, protects the underlying layers of the eye from the external environment and acts as a barrier to pathogens. The corneal epithelium is also responsible for the efficient repair of abrasions. ^1,2 Such repair is highlighted by the ability of corneal epithelial cells to migrate rapidly to the wounded area, with subsequent reepithelialization and reestablishment of barrier function. Impaired wound healing is a major corneal problem evident during certain clinical conditions, including diabetes mellitus. ^{3 –5}  

A variety of changes occur in the microenvironment as corneal epithelial cells migrate to heal wounds. These include alterations in the level of soluble growth factors and their associated signaling receptors, the activity of extracellular proteases, and the reorganization of cell-extracellular matrix (ECM) and cell-cell adhesion sites. ^2,6 For example, corneal perturbations activate the epidermal growth factor receptor (EGFR) and downstream Ras-Raf-Mek-Erk1/2 and PI3K-Akt signaling cascades, which are required for efficient wound healing and are attenuated in patients with diabetic keratopathies. ^{7 –9} Less is known about how cell-cell communication pathways are modulated to allow for directed cell migration into the wounded area. 

Eph receptors constitute a large family of receptor tyrosine kinases that mediate cell-cell communication when engaged by ephrin ligands found on the surfaces of neighboring cells. ^10,11 These receptors are subdivided into EphA and EphB subclasses based on sequence homology and binding preferences for glycosylphosphatidylinositol (GPI)-anchored ephrin-A or transmembrane ephrin-B ligands. Although Eph receptors and ephrins have been most extensively studied in the context of embryonic development and tumor progression, these juxtamembrane signaling complexes are abundantly expressed in adult epithelia. ¹² EphA and EphB receptors and their cognate ligands have also been found in corneal epithelial cells, but their biological function in the anterior ocular surface epithelium remains unknown. ^{13 –15}  

Among the many diverse functions attributed to Eph/ephrin signaling complexes is their ability to regulate epithelial cell migration during tissue morphogenesis, repair, and neoplasia. Interestingly, ligand-dependent activation of EphA receptors is capable of restricting the migration of keratinocytes derived from human skin. ¹⁶ The particular EphA subtypes involved and the mechanisms by which ephrin-A ligands restrain keratinocyte migration have not been directly addressed. However, migration commonly involves the reorganization of integrin-based cell-ECM contacts and the actin-based cytoskeleton by the regulation of src family kinases, focal adhesion kinase, and Rho-family GTPases in many tumor cell lines. ^10,17,18 Notably, EphA2 is capable of dampening Erk1/2 and Akt signaling pathways that are central to corneal wound healing. ^{19 –21}  

Given the prominent expression of EphA2 in epithelial cells and its predilection for ephrin-A1 ligand, which normally limits keratinocyte migration, we sought to determine the distribution and significance of this receptor-ligand pair in the human corneal epithelium. We show that EphA2/ephrin-A1 signaling complexes have the potential to regulate corneal epithelial cell migration and implicate increases in ephrin-A1 expression as part of a pathologic response that targets EphA2 to attenuate Akt signaling and wound healing in persons with diabetes. 

Methods

Cell Culture

The telomerase-immortalized human corneal epithelial cell line hTCEpi was obtained from H. Dwight Cavanaugh and Danielle M. Robertson (University of Texas Southwestern Medical Center, Dallas, TX) and was maintained in keratinocyte–serum-free media (K-SFM; Invitrogen, Carlsbad, CA). ²² Certain experiments were conducted using primary human corneal epithelial keratinocytes (HCEKs) isolated from corneoscleral rims obtained from the Illinois Eye Bank (Chicago, IL) in accordance with the provisions of the Declaration of Helsinki for research involving human tissues. Primary cultures were maintained for up to two passages on collagen IV–coated plates (BD BioSciences, Bedford, MA) in CnT-20 media (Zen Bio Inc., Research Park Triangle, NC), as previously described. ²³  

hTCEpi and HCEKs were incubated in the presence of a soluble recombinant ephrin-A1-Fc peptide (0.01–2.0 μg/mL; R&D Systems, Minneapolis, MN) to stimulate EphA2 receptor activity. Human Fc protein (Jackson ImmunoResearch, West Grove, PA) was used as a negative control. In some experiments, cells were preincubated for 2 hours in the presence of the Mek1/2 inhibitor U0126 (10 μM) or the PI3K inhibitor LY294002 (10 μM; Cell Signaling Technology, Danvers, MA). To study the effects of high glucose on corneal epithelial cells in vitro, hTCEpi cells were also cultured in supplement-free medium containing 25 mM-d-glucose or 8 mM d-glucose with 17 mM mannitol as a control. ⁸  

Gene Expression and Silencing

A full-length human EphA2 cDNA in the retroviral vector, pBABE-puro, was obtained from Bing-Cheng Wang (Case Western Reserve University, Cleveland, OH). ²⁰ A full-length human ephrin-A1 cDNA was obtained from Waldemar Debinski (Wake Forest University Medical Center, Winston-Salem, NC) ²⁴ and subcloned into the pLZRS-Linker vector. ²⁵ A retroviral vector encoding a constitutively active mutant of Raf-1 (pBABE-Raf22W) was provided by Natalia Mitin and Channing Der (University of North Carolina). ²⁶ A constitutively active form of Akt (pBABE-Myr-Akt) was provided by Navdeep Chandel (Northwestern University). ²⁷ Retroviral supernatants were generated and used to transduce hTCEpi cells, as previously described. ²⁸ To silence receptor expression, Stealth siRNA oligonucleotide duplexes (50 nM) targeting EphA2 or a GC-matched negative control (Invitrogen) were transiently transfected into hTCEpi cells using siRNA transfection reagent (Dharmafectin-1; Dharmacon, Lafayette, CO), as described. ²⁹  

In Vitro Scratch Wound Healing Assay

Confluent cultures of hTCEpi cells and HCEKs maintained in growth factor–supplemented medium were used for scratch wound assays as described previously. ²³ Complete culture medium was incubated with the indicated concentration of Fc or ephrin-A1-Fc peptide after scratching for wound healing studies. Experiments were also conducted with hTCEpi cells that had been maintained in supplement-free K-SFM for 72 hours before scratch with the addition of only recombinant peptide after wounding. Images of wound closure were captured at the indicated time points using a digital camera (AxioCam MR; Carl Zeiss, Thornwood, NY) mounted on an inverted light microscope (Axiovert 40 CFL; Carl Zeiss). The percentage of wound closure for each condition was calculated by comparing the cell-free surface area at each time point normalized to the respective wound area at 0 hours using digital image processing software (Axiovision; Carl Zeiss). The scratch wound studies were performed in triplicate for each condition and repeated three separate times for each experiment. All values were expressed as the mean ± SE of all values collected per condition, and the significance between two groups was evaluated by an unpaired Student's t-test. To rule out the contribution of proliferation in the sealing of linear scratch wounds, ephrin-peptide stimulation assays were also performed after treating cultures with 5 μg/mL mitomycin C (EMD BioSciences, San Diego, CA). 

Immunoprecipitation and Western Blot Analysis

Protein lysates were prepared in RIPA buffer (Cell Signaling Technology, Danvers, MA) from intact or wounded hTCEpi cultures treated with Fc or ephrin-A1-Fc at indicated times and subjected to immunoprecipitation or Western blot analysis, as previously described. ³⁰ Immunoprecipitation of EphA2 was carried out from 200 to 300 μg precleared protein lysate using 2 μg/mL rabbit polyclonal anti–EphA2 antibody (C-20; Santa Cruz Biotechnology, Santa Cruz, CA). The C-20 rabbit polyclonal antibody immunoprecipitated EphA2 but not other tyrosine phosphorylated proteins after ephrin-A1-Fc stimulation (Supplementary Fig. S1A). Immune complexes were separated by SDS-PAGE and probed with a mouse monoclonal antibody for phosphotyrosine (4G10; Millipore, Billerica, MA) or EphA2 (D7; Millipore). 

For Western blot analysis, 25 μg of protein lysate was separated by SDS-PAGE and probed with the following primary antibodies: mouse monoclonal antibodies against EphA2 (D7) and phosphotyrosine (4G10); rabbit polyclonal antibodies against ephrin-A1 (V18; Santa Cruz Biotechnology), EphA2 (C-20), EphA3 (C-19; Santa Cruz Biotechnology), EphA4 (S-20; Santa Cruz Biotechnology), total and phospho-Ser473-Akt (Cell Signaling Technology), total and phospho-Thr202/Tyr204-Erk1/2 (Cell Signaling Technology), phospho-Mek1/2 (Cell Signaling Technology), Raf (Cell Signaling Technology), and GAPDH (Calbiochem, San Diego, CA); goat-polyclonal antibodies specific for human EphA2 (AF3035; R&D Systems) and human EphA1 (AF638; R&D Systems). The three EphA2 antibodies (D7, AF3035, and C-20) were tested in Western blot analysis containing lysates prepared from control or EphA2-deficent hTCEpi cells (Supplementary Fig. S1B). 

Immunocytochemistry and Immunohistochemistry

hTCEpi cells were grown on glass coverslips, washed in PBS, and fixed in 4% paraformaldehyde for 5 minutes before permeabilization in 0.05% Triton X-100 for 5 minutes. Immunofluorescence staining was performed as previously described ²⁹ using a mouse monoclonal antibody against EphA2 (D7) or rabbit polyclonal antibody against ephrin-A1 (V18) with detection using an Alexa Fluor-488 or -555 nm–conjugated goat anti–mouse or anti–rabbit antibody (Invitrogen). DAPI was used to counterstain nuclei. Images were acquired using a 40 × 0.5 objective (EC Plan-Neofluar; Carl Zeiss) on an epifluorescence microscope system (AxioVision Z1; Carl Zeiss) fitted with a slide module (Apotome; Carl Zeiss) and a digital camera (AxioCam MRm; Carl Zeiss). 

Frozen sections (5 μm) of OCT-embedded human corneas were fixed in 4% paraformaldehyde, blocked in 2.5% serum, 1% BSA, and 0.05% Tween-20 in PBS, and incubated overnight with primary antibody, goat anti–human EphA2 (AF3035) or rabbit anti–ephrin-A1 (V18). Detection of primary antibodies was carried out using appropriate secondary antibodies conjugated to Alexa-555 or Alexa-488 (Invitrogen). In addition, paraffin-embedded normal (n = 2) and diabetic (n = 5) corneal tissue sections were acquired from the Northwestern University Ophthalmic Pathology Archival Tissue Repository (Chicago, IL). Paraffin-embedded sections were processed for immunohistochemical analysis as previously described. ³⁰ The goat anti–human EphA2 antibody (AF3035) failed to detect this receptor in paraffin-embedded human specimens. Consequently, we used a rabbit anti–EphA2 (C-20) or anti–ephrin-A1 (V18) antibody to immunolocalize this receptor and ligand, respectively, in these archival human tissues. The eyes of wild-type C57/BL6 mice fed a normal diet (n = 4) and a high-fat diet (45 kcal%; n = 6) for 10 weeks (13 weeks of age) were provided by Amy S. Paller (Northwestern University); corneas were isolated and embedded in OCT. Frozen sections of mouse cornea were immunostained using a goat-anti–mouse-EphA2 (AF639; R&D Systems) or anti–ephrin-A1 (V18) antibody. The slides were mounted using DAPI mounting medium (Vector Laboratories, Burlingame, CA) and were observed using a 40 × 0.5 objective (EC Plan-Neofluar; Carl Zeiss) on the epifluorescence microscope system described. ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html) was used to calculate the pixel intensity from the corneal epithelium of at least three separate image fields taken from each individual; these values were normalized for total area to calculate the mean fluorescence intensity. 

Real-Time Quantitative PCR Analysis

The eyes of wild-type C57/BL mice maintained on a normal (n = 3) or a high-fat (n = 3) diet were provided by Amy S. Paller (Northwestern University). Corneal epithelial sheets were isolated by incubating whole globes in PBS containing 20 mM EDTA for 1 hour at 37°C, as described previously. ³¹ Total RNA from these epithelial sheets was harvested using a purification kit (RNeasy; Qiagen, Valencia, CA), and cDNA was prepared using a reverse transcription kit (Superscript III; Invitrogen). Real-time qPCR was performed (7000 Real Time PCR System; Applied Biosystem, Foster City, CA) using a quantitative SYBR green PCR kit (Qiagen), as previously described. ³⁰ Primer sequences used in this study were as follows: mouse EphA2—forward, 5′-ACTGAAAGCGGGCTACACTGAGAA-3′; reverse, 5′-AGGCGGATGATATTGTGGTGGCTA-3′; mouse Ephrin-A1—forward, 5′-GTTTAACCAGCCCAACTGTGCCAT-3′; reverse, 5′-AGGTCCGCACAGCTTGTTTCTTTG-3′. Mouse 18S RNA was used as the internal control. 

Results

EphA2 Receptor Is Primed for Activation by Ephrin-A1 Ligand in Human Corneal Epithelial Cells

The main objective of this study was to determine whether EphA/ephrin-A signaling complexes played a role in corneal epithelial cell migration. We focused on the EphA subfamily of receptors because ephrin-A ligands are capable of restricting epidermal keratinocyte migration. ¹⁶ Moreover, EphA2 is the major subtype that responds to ephrin-A1 for the purpose of dampening Erk1/2-MAPK signaling, limiting growth, and promoting differentiation ^29,32 ; its specific contribution to keratinocyte migration and its role in corneal epithelial cells are not known. 

To begin to study the function of EphA2 in the corneal epithelium, we examined the expression of this receptor and the ephrin-A1 ligand in human cornea, and epithelial cell cultures derived from this tissue. EphA2 was present at cell-cell borders throughout all layers of the corneal epithelium and in hTCEpi cell cultures (Fig. 1A). Although EphA2 and ephrin-A1 were readily detectable in cultured corneal epithelial cells by Western blot analysis (Fig. 1B), ephrin-A1 exhibited a broader distribution throughout these cells in vivo and in vitro (Fig. 1A). Relatively low levels of EphA2 tyrosine phosphorylation were detected in hTCEpi cells under basal conditions, but receptor activation was markedly enhanced by the addition of increasing concentrations of recombinant ephrin-A1-Fc peptide (Fig. 1C). In particular, 0.1 μg/mL ephrin-A1-Fc was sufficient to elevate EphA2 phosphorylation, with maximal activation observed around 1.0 μg/mL of peptide after 15 minutes. These findings demonstrate that EphA2 is expressed on the surfaces of corneal epithelial cells, where it is capable of responding to local increases in ephrin-A ligand. 

Figure 1.

 

EphA2/Ephrin-A1 expression in human corneal epithelial cells in vivo and in vitro. (A) Immunohistochemical analysis of EphA2 (green) and ephrin-A1 (red) in human cornea (left) or hTCEpi cells (right). DAPI (blue) was used to stain nuclei. Nonspecific goat IgG and mouse or rabbit IgG were used as controls for EphA2 and ephrin-A1, respectively. Dotted line: basement membrane zone in vivo. Scale bar, 40 μm. (B) Western blot analysis of EphA2, ephrin-A1, or GAPDH in primary HCEKs or hTCEpi cell lines. (C) hTCEpi cells were stimulated with soluble human Fc peptide (1.0 μg/mL) as a control or increasing concentrations (0.01–2.0 μg/mL) of ephrin-A1-Fc (EfnA1-Fc) peptide for 15 minutes. Protein lysates were harvested for immunoprecipitation (IP; top) or Western blot analysis (bottom). Immunoprecipitated EphA2 was probed for phosphotyrosine (P-Tyr) or total receptor levels. RIPA soluble lysates were also immunoblotted for total EphA2 or GAPDH.

Ligand Activation of EphA2 Restricts the Migration of Corneal Epithelial Cells

Corneal epithelial cells efficiently seal linear scratch wounds in culture, a migratory process that can be inhibited by ephrin-A ligands in epidermal keratinocytes and various cancer cell lines. ^16,20 Thus, we wanted to explore the effects of ephrin-A1 in corneal epithelial cell migration. Basal EphA2 activity remained low in wounded hTCEpi cultures, with a slight increase detectable after 12 hours when cell-free areas had largely been sealed (Fig. 2A). Potentiating EphA2 activity by the addition of ephrin-A1-Fc peptide (1.0 μg/mL) led to near complete loss of the receptor by 6 hours. Coincident with the transient activation and subsequent loss of EphA2 was a marked inhibition in the ability of hTCEpi cells to migrate into the wounded area (Fig. 2B); a similar trend was observed with hTCEpi cells that had been deprived of growth supplement for 72 hours before wounding (data not shown). Inhibition of primary corneal epithelial cell (HCEK) migration was further found after ligand stimulation but with slightly delayed kinetics in wound closure (Fig. 2C). Importantly, overnight pretreatment with mitomycin C did not interfere with the ability of ephrin-A1-Fc to restrict the sealing of wounds (data not shown), indicating that a decrease in migration, not proliferation, was responsible for the attenuated wound healing response. 

Figure 2.

 

Soluble ephrin ligand activation of EphA2 restricts corneal epithelial cell migration. (A) IP and P-Tyr analysis of EphA2 in hTCEpi cells stimulated with 1.0 μg/mL Fc or EfnA1-Fc for indicated times after generation of a linear scratch wound. GAPDH was used as a reference loading control. Photomicrographs (left) and bar graphs (right) depicting the extent of wound closure for hTCEPi (B) or primary HCEKs (C) in the presence of Fc or EfnA1-Fc peptide is shown from a representative experiment that was performed in triplicate and was conducted on three separate occasions. *P < 0.05.

To determine whether EphA2 was specifically required for the ephrin-mediated restriction in migration, we silenced the expression of this receptor (Fig. 3A). Despite the presence of other related receptors that can bind this ligand, ephrin-A1-Fc peptide treatment in EphA2-deficient hTCEpi cells had little effect on migration beyond the level of the Fc control (Fig. 3B). In addition, EphA2 silencing by itself did not profoundly alter hTCEpi migration at any of the time points examined, indicating that simply downregulating this receptor is not sufficient to enhance migration in vitro (Fig. 3C). These observations suggest that EphA2 is primarily responsible for corneal epithelial cells responding to elevated ephrin-A1 stimulation and limiting the rate of wound healing. 

Figure 3.

 

EphA2 is required for the ligand-induced restriction of migration in hTCEpi cells. (A) Expression of EphA receptors in hTCEpi cells transiently transfected with siRNA oligonucleotide duplexes for EphA2 (siEphA2) or a control siRNA (siCtrl). (B) Scratch wound assay on confluent sheets of control or EphA2-deficient hTCEpi cells stimulated with 1.0 μg/mL Fc or EfnA1-Fc for 6 hours. (C) Scratch assays on confluent sheets of control or EphA2 knockdown hTCEpi cells for 3, 6, 9, and 12 hours after wounding. Data from three independent experiments are represented in the bar graphs. *P < 0.05.

Enhancing EphA2/Ephrin-A1 Signaling Complexes Delays Corneal Epithelial Wound Healing In Vitro

EphA2 is commonly found in tumor cell lines, in which it is thought to promote invasion until confronted by ephrin ligands in the tumor microenvironment. ^17,33 In support of this notion, EphA2 overexpression increases the migration of glioma and prostate cancer cell lines that lack ephrin-A1 ligand. ²⁰ Because hTCEpi cell lines express ephrin-A1, we reasoned that elevating the level of EphA2 might instead decrease migration in these untransformed epithelial cell lines by engaging endogenous ligand. Accordingly, ectopic overexpression of EphA2 markedly increased basal receptor tyrosine phosphorylation, which was further enhanced by the addition of soluble ligand after 15 minutes but not after 6 hours (Fig. 4A). At this time, hTCEpi cells overexpressing EphA2 exhibited slower wound healing responses (30.7% ± 3.5%) compared with empty vector (pBABE) controls treated with Fc peptide (48.6% ± 3.8%; P = 0.0049). hTCEpi cells simply overexpressing EphA2 were nearly as restricted in their migration capacity as control cells treated with ephrin-A1-Fc (22% ± 5.2%; P = 0.098; Fig. 4B). The migration of EphA2-overexpressing hTCEpi cells could be further inhibited by the addition of soluble ephrin-A1-Fc ligand (20.2% ± 3.9%; P = 0.036), albeit to a lesser extent than control cells treated with this peptide ligand; these findings demonstrated that elevated EphA2 signaling can restrict migration and further suggested that the effects of EphA2 phosphorylation on migration can be saturated in hTCEpi cells. 

Figure 4.

 

Elevated EphA2 activity inhibits corneal epithelial cell migration. (A) hTCEpi cells transduced with an empty or an EphA2 (hEphA2)–containing pBABE retrovirus were harvested for IP and p-Tyr analysis (top) with or without stimulation by 1.0 μg/mL EfnA1-Fc for 15 minutes or 6 hours. Total EphA2 or GAPDH levels are shown (bottom). (B) Scratch wound assay on confluent sheets of control or EphA2-overexpressing hTCEpi cells stimulated with 1.0 μg/mL Fc or EfnA1-Fc for 6 hours. Data from three independent experiments are represented in the bar graphs. *P < 0.05.

Furthermore, we predicted that introducing more ligand into hTCEpi cells would engage endogenous EphA2 and restrict migration. Ectopic ephrin-A1 expression led to a marked reduction in EphA2 levels, as would be expected from increased signaling, ²⁴ which was confirmed biochemically (Fig. 5A) and was reflected by the concentration of ephrin-A1 ligand and the depletion of EphA2 receptor at cell-cell contacts by immunostaining (Fig. 5B). Importantly, elevated ephrin-A1 was capable of inhibiting corneal epithelial cell migration in a manner that no longer depended on the addition of soluble ligand (Fig. 5C). Collectively, these observations showed that modulating the levels of EphA2 and ephrin-A1 can influence corneal epithelial migration and further suggested that pathologic conditions that lead to increased ligand expression may impair wound healing. 

Figure 5.

 

Increased ephrin-A1 expression inhibits corneal epithelial cell migration. (A) hTCEpi cells transduced with an empty or an ephrin-A1 (hEfnA1)–containing pLZRS retrovirus were harvested for EphA2 IP and p-Tyr analysis (top). Total EphA2, ephrin-A1, and GAPDH levels are also shown (bottom). (B) Immunofluorescence analysis of EphA2 (green) and ephrin-A1 (red) on confluent sheets of control or ephrin-A1–overexpressing hTCEpi cells. Scale bar, 40 μm. (C) Scratch wound assay on confluent sheets of control or ephrin-A1–overexpressing hTCEpi cells stimulated with 1.0 μg/mL Fc or EfnA1-Fc for 6 hours. Data from three independent experiments are represented in the bar graphs. *P < 0.05.

Ephrin-A1 Ligand-Mediated Restriction in Cell Migration Involves Suppression of the Akt Pathway

Ligand activation of EphA2 inhibits Erk1/2-MAPK and PI3K-Akt signaling ^{19 –21} ; these pathways are key regulators of cell proliferation and survival but also promote corneal epithelial cell migration. ^7,8 Under steady state conditions, hTCEpi cells lacking EphA2 or overexpressing this receptor or its ephrin-A1 ligand showed no gross changes in the basal activation state of either Erk1/2 or Akt (Supplementary Fig. S2). In contrast, ephrin-A1-Fc delivery was capable of dampening Erk1/2 and Akt activation in hTCEpi cells in a dose- and time-dependent manner, coincident with the inhibition in migration (Figs. 6A, 6B). In particular, 0.1 μg/mL ephrin-A1-Fc was sufficient to inhibit migration and Akt signaling, whereas marked inhibition of the Erk1/2-MAPK pathway was observed only at higher concentrations (≥0.5 μg/mL) of the peptide. The lower level of receptor activation induced by 0.1μg/mL ephrin-A1-Fc inhibited migration to an extent similar to that observed with 1.0 μg/mL concentration of the peptide (Fig. 6C), where EphA2 phosphorylation was at its maximum (Fig. 1C). These observations provided more evidence that EphA2 reaches a threshold of activation to impact migration and further suggested that ligand activation of EphA2 may only have to suppress Akt and not Erk1/2 signaling to attenuate wound healing responses. 

Figure 6.

 

Ligand activation of EphA2 restricts corneal epithelial cell migration by suppressing Akt. Immunoblot analysis of phosphorylated or total Akt (Ser473) and Erk1/2 (Thr202/Tyr204) in hTCEpi cells stimulated with (A) 1.0 μg/mL Fc or increasing concentrations of EfnA1-Fc for 6 hours; (B) 1.0 μg/mL Fc or EfnA1-Fc for indicated times. (C) Scratch wound assay on confluent sheets of hTCEpi cells treated with 1.0 μg/mL Fc or increasing concentrations of EfnA1-Fc. (D) Scratch wound assays were performed after 2-hour pretreatment with 10 μM LY294002 or 10 μM U0126 to inhibit the PI3K-Akt or Mek-Erk1/2 pathway, respectively. *P < 0.05 comparing EfnA1-Fc and Fc. #P < 0.05 comparing the Fc of each inhibitor with the vehicle control. (E) Scratch wound assays were performed on growth factor–depleted confluent sheets of control (pBABE) or Raf22W-overexpressing hTCEpi cells stimulated with 1.0 μg/mL Fc or EfnA1-Fc for 12 hours. Immunoblot analysis of Raf, phosphorylated Mek1/2, or Erk1/2 (Thr202/Tyr204) in hTCEpi cells stimulated with 1.0 μg/mL Fc or EfnA1-Fc for 12 hours is shown next to the bar graphs. (F) Scratch wound assays were also performed on confluent sheets of control (pBABE) or Myr-Akt–overexpressing hTCEpi cells maintained in complete culture medium, wounded, and then stimulated with 1.0 μg/mL Fc or EfnA1-Fc for 6 hours. Immunoblot analysis of phosphorylated or total Akt (Ser473) at 6 hours is shown next to the bar graph.

To determine whether ephrin-A1 ligand can restrict hTCEpi motility in the absence of Erk1/2 or Akt signaling, we performed scratch wound assays after pharmacologic blockade of the upstream activator Mek1/2 (U0126) or PI3K (LY294002). As expected, interfering with Erk1/2 or Akt signaling significantly inhibited wound closure in Fc-treated control cultures (Fig. 6D). Although Erk1/2 inhibition most profoundly delayed wound healing, the addition of ephrin-A1-Fc peptide further exacerbated this defect. In addition, the introduction of a constitutively-active Raf mutant (Raf22W ²⁶ ) to elevate Mek-Erk1/2 signaling did not prohibit the ability of soluble ephrins in delaying wound healing, demonstrating that ligand activation of EphA2 most likely restricts corneal epithelial cell migration in an Erk1/2-independent manner (Fig. 6E). 

In contrast, ephrin-A1 ligand delivery had no residual effect on wound closure in PI3K-inhibited cultures (Fig. 6D). In light of these observations, we used a myristoylated variant of Akt (Myr-Akt) that targets this protein to lipid membranes, thereby increasing its activation. ²⁷ Elevated Akt signaling in Fc control cells was confirmed biochemically and increased basal migration compared with an empty vector control (Fig. 6F). Interestingly, ephrin-A1-Fc ligand stimulation remained capable of dampening Akt signaling in cells expressing the Myr-Akt mutant, consistent with previous reports. ³⁴ Accordingly, ephrin-A1-Fc peptide treatment suppressed migration compared with Fc controls in these cells (Fig. 6F). These observations are consistent with the idea that dampening of the Akt signaling pathway is involved in the ephrin-mediated response that limits corneal epithelial cell migration. 

Elevated Ephrin-A1 Levels in Response to High Glucose Treatment and in Diabetic Corneal Epithelium

Delayed wound healing responses in the diabetic cornea are associated with reduced Erk1/2 and Akt signaling, which can be partially mimicked in vitro by growing corneal epithelial cells in the presence of high glucose levels. ^7,8 Given that ligand activation of EphA2 attenuates Akt signaling and corneal epithelial migration, we wanted to test whether these changes would be apparent under high-glucose conditions. Accordingly, high-glucose exposure resulted in an increase in the expression of ephrin-A1, which was associated with a corresponding decrease in EphA2 levels, reduced Akt and Erk1/2 phosphorylation, and decreased migration (Figs. 7A, 7B). 

Figure 7.

 

Ephrin-A1 is increased in response to high-glucose and diabetic corneas. (A) Immunoblot analysis of EphA2, ephrin-A1, phosphorylated or total Akt (Ser473), and Erk1/2 (Thr202/Tyr204) in hTCEpi cells cultured in supplement-free medium containing 25 mM-d-glucose (G) or 8 mM d-glucose with 17 mM mannitol as a control (M) for 72 hours. (B) Time course of scratch wound assay on confluent sheets of control (M) or high glucose–treated hTCEpi cells. Data from three independent experiments are represented in the bar graphs. *P < 0.05. (C) Immunofluorescence analysis of ephrin-A1 expression in healthy (top) and diabetic (bottom) human (left) and mouse (right) corneas from animals fed a high-fat diet (HFD). Dotted line: Basement membrane zone. Scale bar, 40 μm. (D) Bar graphs represent the fold change in mean fluorescence intensity (MFI) of five diabetic corneas compared with two individual controls (left) and four control mouse corneas on a normal diet (ND) compared with six diabetic mice fed a HFD (right). Error bars (±SE) reflect the variation in pixel intensity from multiple fields (n > 3) acquired for each cornea. (E) Bar graphs represent qPCR analysis of ephrin-A1 mRNA levels from corneas in ND and diabetic mice fed with a HFD (n = 3) normalized to 18S RNA levels.

Based on these in vitro findings, we asked whether ephrin-A1 levels were altered in diabetic corneal tissues. Indeed, a marked increase in ephrin-A1 immunoreactivity was observed in the corneal epithelium of individuals with diabetes mellitus (n = 5; Fig. 7C). We could not detect peripheral staining of EphA2 in paraffin-embedded sections of human cornea using the three anti–EphA2 antibodies tested in this study. A slight reduction in immunostaining of EphA2 was observed using a rabbit polyclonal antibody for this receptor that exhibits a distinct cytoplasmic distribution in corneal epithelium (Supplementary Fig. S3A); however, it remains unclear whether the additional protein species recognized by this antibody in EphA2-deficient hTCEpi cells contributes to these differences (Supplementary Fig. S1B). 

To confirm these findings in an in vivo model of the diabetic state, we examined ephrin-A1 and EphA2 immunoreactivity and mRNA levels in the corneas of mice maintained on a normal diet compared with animals fed a high-fat diet for 13 weeks, as previously described. ^35,36 Although qRT-PCR showed no significant changes in ephrin-A1 mRNA levels (Fig. 7E), we detected an increase in ephrin-A1 immunoreactivity in corneas isolated from mice on a high-fat diet (Figs. 7C, 7D). In addition, EphA2 mRNA levels and immunoreactivity were not altered in the diabetic mouse cornea (Supplementary Fig. S3). These observations indicate that ephrin-A1 is likely upregulated by posttranscriptional mechanisms in the corneal epithelium of diabetic individuals and in these two experimental models of diabetes; stabilization of this ligand on the surfaces of corneal epithelial cells may provide an inhibitory cue for EphA2 in the cellular microenvironment that impairs Akt-mediated wound healing responses. 

Discussion

EphA2 plays a key role in ocular biology because inherited mutation or targeted gene deletion of this receptor is associated with cataract development. ^{37 –39} These defects in lens clarity are related to aberrant function of EphA2 or one of its ligands, ephrin-A5, in lens epithelial and fiber cells. ^{37 –40} We now show that EphA2 may also mediate signaling events in the ocular anterior segmental epithelia, specifically an involvement of EphA2/ephrin-A1 signaling complexes in corneal epithelial wound healing. We also demonstrate that the suppression of the PI3K-Akt pathway by EphA2 is a means to inhibit corneal epithelial cell motility. 

EphA2 was localized to the periphery of cells in the human corneal epithelium at sites where it is capable of responding to ephrin ligands and mediating cell-cell communication events. We also detected EphA2 in corneal epithelial cell cultures, along with multiple related receptors and the ephrin-A1 ligand. Although EphA2 was previously found in immortalized mouse corneal epithelial cells, this receptor was not detectable by immunohistochemical means in mouse cornea in vivo; instead, EphA3 appeared to be the prominent EphA subtype in the murine corneal epithelium. ¹³ It is possible that species-specific gene expression patterns account for these differences in corneal EphA receptor distribution, but our immunodetection of EphA2 in human cornea is in accordance with its expression in other stratified epithelia and microarray profiles deposited in the Gene Expression Omnibus databank ^32,41 and with our qPCR analysis in wild-type C57/BL6 mouse corneas (Supplementary Fig. S3C). We have not specifically addressed the role of EphA3 in this study. However, ephrin-A1 ligand failed to restrain the migration of corneal epithelial cells lacking EphA2 despite the persistent expression of EphA3 in these cultures. These finding suggest that EphA2 is the predominant receptor subtype that mediates corneal epithelial wound healing responses to ephrin-A1 ligand stimulation, at least in vitro. 

EphA2 has previously been implicated in cell migration, but this has most often been studied in the context of tumors, in which it is frequently overexpressed. Under these circumstances, EphA2 promotes cancer cell invasion when it is not engaged by ephrin ligands in a manner that involves Akt-mediated phosphorylation of a serine residue (Ser897) on its cytoplasmic domain. ^20,42 In contrast, ligand-dependent activation of EphA2 leads to receptor internalization and degradation, suppression of Akt signaling, and ultimately inhibition of tumor cell motility. EphA2 does not appear to serve as an essential effector of Akt-mediated motility in corneal epithelial cells in steady state conditions because the simple loss of this receptor does not significantly alter wound healing responses. Instead, ephrin-mediated inhibition of Akt by EphA2 acts to limit cell migration. The relative abundance of this GPI-linked ligand on the surface of the neighboring cell might determine the migratory fate of the corneal epithelium expressing EphA2 because ectopically increasing the levels of ephrin-A1 led to its concentration at cell-cell contacts and further restricted migration (Fig. 5). 

Inhibition of Akt by EphA2 is mediated by a PP1-like serine/threonine phosphatase in prostate cancer cell lines, with mutations in the PTEN lipid phosphatase. ³⁴ It will be of interest to determine whether this serine/threonine phosphatase or other upstream regulators of the Akt pathway are required for EphA2-mediated inhibition of motility in untransformed corneal epithelial cells with intact PTEN. In fact, ligand activation of EphA2 was still capable of dampening the phosphorylation of the Myr-Akt mutant (Fig. 6F). Increased phosphatase activity can presumably continue to inactivate membrane-associated Akt under these conditions, consistent with a previous study in which soluble ephrin ligands inhibited Myr-Akt phosphorylation levels ³⁴ ; this might explain how ephrin-A1-Fc still inhibits migration in corneal epithelial cells expressing the Myr-Akt mutant. 

The activation of EGFR and downstream signaling through the Erk1/2-MAPK and PI3K-Akt pathways during corneal epithelial wound healing has been studied extensively. Interestingly, these signaling pathways are dampened in human and rodent diabetic corneas in which wound healing is delayed. ^7,8 An increase in reactive oxygen species induced by elevated glucose ⁸ and the overexpression of proteinases, such as matrix metalloproteinase-10 and cathepsin F, ^43,44 are capable of reducing Akt signaling, but other signaling mechanisms may exist to dampen pathways downstream of EGFR in this disease. The increase in ephrin-A1 immunoreactivity observed in diabetic corneas may serve this function because ligand activation of EphA2 leads to a diminution of Akt signaling. Alternatively, EphA2 may directly interfere with EGFR by forming a complex with this receptor ⁴⁵ or enhancing its degradation by the Cbl ubiquitinase. ^{46 –48} Moreover, EphA2 is required for the normal suppression of Erk1/2-MAPK signaling in epidermal keratinocytes ^29,32 and may also regulate this pathway in corneal epithelial cells through p120RasGAP. ^49,50 We provide evidence that EphA2 levels are reduced in corneal epithelial cells maintained in high glucose (Fig. 7A) and may be downregulated in the corneal epithelium of diabetic individuals (Supplementary Fig. S3), which is consistent with the presence of additional ephrin-A1 ligand. Receptor levels may also decrease as a result of lower Erk1/2 activity because EphA2 gene expression itself is directly regulated by an EGFR-Erk1/2-MAPK pathway. ^19,45,51  

Pathologic alterations in the Eph/ephrin axis may occur in other ocular diseases that affect cell motility in the anterior segmental epithelium beyond diabetic keratopathies. For example, pterygium is characterized by the overgrowth of conjunctival and limbal epithelial cells in the direction of the cornea and is commonly associated with an upregulation of genes associated with cell migration and elevated Erk1/2-MAPK signaling. ^52,53 Furthermore, ephrin-A1 mRNA transcripts are reduced in these fibrovascular lesions compared with the unaffected conjunctival epithelium. ⁵⁴ It will be interesting to determine whether the downregulation of ephrin-A1 represents a loss of an important inhibitory cue for conjunctival or limbal epithelial cells that leads to their invasion into the cornea. Similarly, IL-1 stimulation of corneal epithelial cell cultures reduces the levels of the related ephrin-A5, suggesting a broader role for these ligands in corneal inflammatory and wound healing responses. ^14,55 These findings raise the possibility that the EphA2/ephrin-A axis helps maintain normal tissue compartmentalization in the conjunctiva, limbus, and cornea by preventing the migration of epithelial cells into fields with high levels of ephrin ligand. 

Supplementary Materials

Text s1, DOC - Text s1, DOC 

Figure sf01, TIF - Figure sf01, TIF 

Figure sf02, TIF - Figure sf02, TIF 

Figure sf03, TIF - Figure sf03, TIF 

Footnotes

 Supported in part by the Dermatology Foundation, Zell Family Foundation, and Foglia Family Foundation (SG); National Institutes of Health/National Eye Institute Grants EY06769, EY017536, and EY019463 (RML); Research to Prevent Blindness (PB); and National Institute of Arthritis and Musculoskeletal and Skin Diseases Grant AR057216 (NU-SDRC Pathology Core facility).

Footnotes

 Disclosure: N. Kaplan, None; A. Fatima, None; H. Peng, None; P.J. Bryar, None; R.M. Lavker, None; S. Getsios, None

The authors thank Bing-Cheng Wang (Case Western Reserve University), Waldemer Debinski (Wake Forest University), Natalia Mitin and Channing Der (University of North Carolina), and Navdeep Chandel (Northwestern University) for the EphA2, ephrin-A1, Raf22W, and Myr-Akt cDNA constructs; Danielle M. Robertson and H. Dwight Cavanaugh (University of Texas Southwestern Medical Center) for the hTCEpi cell line; Amy S. Paller (Northwestern University) for mouse eyes; and the staff of the NU-SDRC Pathology Core facility for assisting in morphologic analyses.